156
157
RBM5 was originally identified as a putative tumor suppressor gene consistent with the frequent deletion of its gene locus in lung cancer; while it is over-expressed in breast cancer, suggesting a putative role in promoting tumorigenesis. Therefore, it is very interesting to study certain RBM5 point mutations which are reported in certain cancer patients. To this end, I used cBioPortal (http://www.cbioportal.org/), a portal for cancer genomics that provides visualization, analysis and downloading of large-scale cancer genomics data. It has a nice mutation assessor feature, which can be used to assess the functional impact of the mutation based on conservation. For RBM5, 132 missense mutations are listed. I selected three point mutations based on the functional impact score and the position of the mutation (in the RNA binding domains). For RRM1, I chose R115P and R140S point mutations which are scored to have a ‘high’ functional impact on the protein via the mutation assessor. For RRM2, I chose R263H point mutation which was ranked to have a ‘medium’ functional impact. The reason for choosing this particular mutation was that another point mutation in the same residue but to a proline (R263P) was reported previously to cause male sterility in mice (O'Bryan, Clark et al. 2013).
It was suggested that R263P mutation causes pre-mRNA splicing defects in a number of its targets in mice testis, thereby leading to sterility. The authors postulated that since R263 is present on the -sheet surface and possibly involved in RNA binding, its mutation to a proline might have two-fold effects- disturbance of local secondary structure and abrogation of RNA interaction as a direct effect of substitution of the RNA binding residue R263 to a proline residue. I was therefore interested in studying which of the aforementioned effects is a direct reason of pre-mRNA splicing defects of the mutant protein.
Firstly, the two proline mutations (R115P in RRM1 and R263P in RRM2) were cloned in the single domains, expressed in M9 minimal medium and purified as before. Surprisingly, the mutant proteins did not purify as well as the wild-type proteins, which was already an indication that the mutants had major effects on the proteins. Nevertheless, the purification was still successful and it was possible to record 1H,15N-HSQC spectra of the mutant proteins. In Figure 70, a superposition of 1H,15N-HSQC spectra of wild-type RRM1 (blue) and R115P mutant (orange) proteins is shown in panel A while that of wild-type RRM2 (pink) and R263P mutant (grey) proteins is shown in panel B. Comparison of the wild-type and mutant spectra clearly shows that in both the cases, the proline mutation disrupts the structure of the protein.
Although RRM1 R115P mutant protein is partially unstructured, RRM2 R263P mutant protein is completely unstructured.
158
Figure 70 Disease linked mutations affecting the secondary structure of the domains
Superposition of 1H,15N -HQSC spectra of type RRM1, RRM1 R115P mutant protein and wild-type RRM2, RRM1 R263P mutant proteins in blue, orange, pink and grey, respectively is shown in panels (A) and (B). The respective position of mutations are shown on the structure of RRM1 and RRM2 (PDB ID: 2LKZ). Both the mutations compromise the structural integrity of the individual domains
.
This suggests that the folding defect in RRM2 R263P mutant is directly translated into functional defects whereby pre-mRNA splicing of its target proteins is affected. Similarly, it is quite possible that the R115P mutation in RRM1 also leads to functional defects due to partially unstructured domain. It is noteworthy that this is just one of the many point mutations in just one protein in the cancer patient and therefore might just be a small contributing factor in causing the disease.
Next, the other two point mutations (R140S in RRM1 and R263P in RRM2) were also cloned and expressed as before. Purification of these two mutants is straightforward, unlike the proline mutants. The proteins also seemed folded upon recording their 1H,15N-HSQC spectra (right panels in Figure 71A, Figure 72A).
159
Figure 71 RRM1 R140S cancer mutation does not affect the structure or RNA binding
(A) Overlays of 1H,15N-HSQC spectra of wild type RRM1 in its free and RNA bound form and RRM1 R140S cancer mutant in its free and RNA bound form are shown in the left and right panels in blue, dark grey, green and maroon, respectively. (B) A comparison of the chemical shift perturbation plots of the wild type RRM1 (dark grey) and R140S mutant (maroon) when bound to the same RNA oligo –CU_9 (5’-UCUCUUCUC-3’) in 1:1 ratio is shown. The position of the mutated residue is shown on the structure of RRM1.
160
Figure 72 RRM2 R263H cancer mutation does not affect the structure or RNA binding
(A) Overlays of 1H,15N-HSQC spectra of wild type RRM2 in its free and RNA bound form and RRM2 R263H cancer mutant in its free and RNA bound form are shown in the left and right panels in pink, dark grey, purple and maroon, respectively. (B) A comparison of the chemical shift perturbation plots of the wild type RRM2 (dark grey) and R263H mutant (maroon) when bound to the same RNA oligo –CU_9 (5’-UCUCUUCUC-3’) in 1:1 ratio is shown. The position of the mutated residue is shown on the structure of RRM2 (PDB ID: 2LKZ).
Since the mutations did not affect the fold, I titrated the Caspase-2 derived RNA oligo CU_9 (5’-UCUCUUCUC-3’) to check if the respective mutations affect RNA binding. No significant changes in the RNA binding pattern is observed in the mutants upon comparison of the chemical shift perturbation plots of the mutant and wild-type proteins titrated with RNA (Figure 71B, Figure 72B). This is not surprising as these are just single point mutations that we are looking at in isolation. It is possible that either these mutations are present in the cancer patient, either by chance or have some other affects (for example on protein-protein interactions etc.), study of which is beyond the scope of this thesis.
161